linear algebra in a computational setting alan kaylor cline deans scholars seminar january 22, 2014

Post on 19-Jan-2018

214 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

After examining the code you believe that the running time depends entirely upon some input parameter n and … a good model for the running time is Time(n) = a + b·log 2 (n) + c·n + d·n·log 2 (n) where a, b, c, and d are constants but currently unknown.

TRANSCRIPT

Linear Algebrain a

Computational Setting

Alan Kaylor Cline

Dean’s Scholars SeminarJanuary 22, 2014

How long does it take for this code to run?

After examining the code you believe that the running time depends entirely upon some input

parameter n and …

a good model for the running time is

Time(n) = a + b·log2(n) + c·n + d·n·log2(n)

where a, b, c, and d are constants but currently unknown.

So you time the code for 4 values of n, namely n = 10, 100, 500, and 1000

and you get the times

According to the model you then have 4 equations in the 4 unknowns a, b, c, and d:

a + b·log2(10) + c·10 + d·10·log2(10) = 0.685a + b·log2(100) + c·100 + d·100·log2(100) = 7.247a + b·log2(500) + c·5000 + d·500·log2(500) = 38.511a + b·log2(1000) + c·1000+ d·1000·log2(1000) = 79.134

Time(10) = 0.685 ms.Time(100) = 7.247ms.Time(500) = 38.511ms.Time(1000) = 79.134 ms.

These equations are linear in the unknowns a, b, c, and d.

a + b·log2(10) + c·10 + d·10·log2(10) = 0.685a + b·log2(100) + c·100 + d·100·log2(100) = 7.247a + b·log2(500) + c·5000 + d·500·log2(500) = 38.511a + b·log2(1000) + c·1000+ d·1000·log2(1000) = 79.134

We solve them and obtain: a = 6.5

b = 10.3 c = 57.1

d = 2.2

So the final model for the running time is

Time(n) = 6.5 + 10.3·log2(n) + 57.1·n + 2.2·n·log2(n)

and now we may apply the model Time(n) = 6.5 + 10.3·log2(n) + 57.1·n + 2.2·n·log2(n)

for a particular value of n (for example, n = 10,000)to estimate a running time of

Time(10,000) = 6.5 + 10.3·log2(10,000) + 57.1· 10,000 + 2.2· 10,000 ·log2(10,000) = 863.47 ms.

Picture from MarsOct. 1, 1976

One approach to the coloring of the images was to adjust manually the colors in the various patches.

Picture from MarsOct. 1, 1976

One approach to the coloring of the images was to adjust manually the colors in the various patches.

Picture from MarsOct. 1, 1976

One approach to the coloring of the images was to adjust manually the colors in the various patches. This is what resulted:

Picture from MarsOct. 1, 1976

The other approach was to solve 6 x 6 linear systems for each pixel. The systems were based on the filters being used.

Picture from MarsOct. 1, 1976

The other approach was to solve 6 x 6 linear systems for each pixel. The systems were based on the filters being used.This is what resulted:

Picture from MarsOct. 1, 1976

Compare the two:

But have we ignored something?

0 100 200 300 400 500 600 700 800 9000

2

4

6

8

10

12

14

16

18x 10

5

So you time the code for 30 values of n, and you get these times {(ni,ti)}

If the model was perfect and there were no errors in the timings

then for some values a, b, c, d, and e:

a + b·log2(ni) + c·ni + d·ni·log2(ni) +e·ni2 = ti

for i =1,…,30

But the model was not perfect and there were error in the timings

So we do not expect to get any values a, b, c, d, and e so that:

a + b·log2(ni) + c·ni + d·ni·log2(ni) +e·ni2 = ti

for i =1,…,30

We will settle for values a, b, c, d, and e so that:

a + b·log2(ni) + c·ni + d·ni·log2(ni) +e·ni2 ti

for i =1,…,30

Our sense of

a + b·log2(ni) + c·ni + d·ni·log2(ni) +e·ni2 ti

for i =1,…,30

Will be to get a, b, c, d, and e so that sum of squares of all of the differences

(a + b·log2(ni) + c·ni + d·ni·log2(ni) +e·ni2 - ti)2

is minimized over all possible choices of a, b, c, d, and e

After solving the least squares system to get the best values of a, b, c, d, and e, we plot

a + b·log2(n) + c·n + d·n·log2(n) + e·n2

0 100 200 300 400 500 600 700 800 900 10000

0.5

1

1.5

2

2.5x 10

6

What’s a “good” solutionwhen we don’t have the exact

solution?

What’s a “good” solutionwhen we don’t have the exact

solution?

“Hey. That’s not a question that was discussed in other math classes.”

What’s a “good” solutionwhen we don’t have the exact

solution?

.780 𝑥+ .563 𝑦= .217

.913 𝑥+ .659 𝑦=.254

Consider the two equations:

Consider two approximate solution pairs:

.780 𝑥+ .563 𝑦= .217

.913 𝑥+ .659 𝑦=.254

and these two equations:

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Consider two approximate solution pairs:

.780 𝑥+ .563 𝑦= .217

.913 𝑥+ .659 𝑦=.254

and these two equations:

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Which pair of these two is better?

Important fact to consider:

𝑥=1𝑦=−1

The exact solution is:

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Which pair of these two is better?

Consider two approximate solution pairs:

.780 𝑥+ .563 𝑦= .217

.913 𝑥+ .659 𝑦=.254

and these two equations:

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Which pair of these two is better?

Important fact to consider:

.780 𝑥+ .563 𝑦= .217

.913 𝑥+ .659 𝑦=.254

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Which pair of these two is better?

.780 𝑥1+.563 𝑦1=.215757

.913 𝑥1+.659 𝑦 1= .252428

.780 𝑥2+.563 𝑦 2=.216999.913𝑥 2+.659 𝑦 2=.254

Recall we are trying to solve:

For the first pair, we have:

For the second pair, we have:

Important fact to consider:

𝑥1𝑦 1

¿¿.999−1.001

𝑥2𝑦 2

¿¿.341−0.087

Which pair of these two is better?

Student: “Is there something funny about

that problem?”

Student: “Is there something funny about

that problem?”

Professor: “You bet your life. It looks innocent but it

is very strange. The problem is knowing when you have a strange case on

your hands.”

CLINE

Professor: “Geometrically, solving equations is like

finding the intersections of lines.”CLINE

here’s the intersection?

When lines have no thickness …

where’s the intersection?

… but when lines have thickness …

25.96 miles

Galveston Island

25.96 miles

Galveston Island

Where’s the intersection?

London Olympics Swimming

• http://www.youtube.com/watch?v=fFiV4ymEDfY&feature=related

• 1:19

How do you transform this image …

How do you transform this image …

into the coordinate system of another image?

and in greater generality, transform

3-dimensional objects

The $25 Billion Eigenvector

How does Google do Pagerank?

The $25 Billion Eigenvector

How did Google do Pagerank?

The Imaginary Web Surfer:

• Starts at any page,• Randomly goes to a page linked from the

current page,• Randomly goes to any web page from a

dangling page,• … except sometimes (e.g. 15% of the time),

goes to a purely random page.

A tiny web: who should get the highest rank?

J A B

I C

DH

G F E

The associated stochastic matrix:

0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.4400 0.0150 0.0150 0.2983 0.4400 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.2983 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.2983 0.8650 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.4400 0.0150 0.0150 0.8650 0.0150 0.8650 0.0150 0.0150 0.0150 0.0150 0.0150 0.2983 0.0150 0.0150 0.8650 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.8650 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.8650 0.2983 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.2983 0.0150 0.0150 0.0150 0.0150 0.0150 0.0150 0.4400 0.0150 0.0150 0.0150

We seek to find a vector x so that

A x = x

One way is to start with some initial x0,and then:

for k = 1, 2, 3,…xk = A xk-1

This converges to an x so that A x = x

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Start with equal components

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

One iteration

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Two iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Three iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Four iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Five iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Six iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Seven iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Eight iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Nine iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Ten iterations

1 20

0.05

0.1

0.15

0.2

0.25

0.3

0.35

The Eigenvector

[U,G] = surfer ('http://cns.utexas.edu/honors/honors-programs-center/deans-scholars', 100)

0 20 40 60 80 100

0

10

20

30

40

50

60

70

80

90

100

Pagerank Power Iteration the limit

0 20 40 60 80 100 1200

0.02

0.04

0.06

0.08

0.1

0.12

0.14Page Rank

And the winners are…

'http://www.utexas.edu''http://www.utexas.edu/maps''http://www''http://m.utexas.edu''http://www.utexas.edu/emergency''http://www.lib.utexas.edu' 'http://www.utexas.edu/parking/transportation/shuttle''http://healthyhorns.utexas.edu''http://www.utexas.edu/know/feed''http://www.utexas.edu/know''http://cns.utexas.edu'

How much storage to hold this array?

• Current estimate of indexed WWW:4.7 · 1010 web pages

• If placed into an array this would have2.21 · 1021 elements

• If each element is stored in 4 bytes, this would be 8.8 · 1022 bytes

• Current estimate of world’s data storage capacity is 3.0 · 1018 bytes (.003% of necessary space)

http://www.smartplanet.com/blog/thinking-tech/what-is-the-worlds-data-storage-capacity/6256

How much time to do one power step?

• Current estimate of indexed WWW:4.7 · 1010 web pages

• If placed into an array this would have2.21 · 1021 elements

• Fastest current machine does 33.86 · 1015

operations per second• One step of y = Ay takes 3.68 days

Go Home Now.

top related